|Home | About | Journals | Submit | Contact Us | Français|
Accumulation of aggregated amyloid-β peptide (Aβ) has been studied as an initial step of Alzheimer’s disease (AD) pathogenesis. Following amyloid plaque formation, reactive microglia and astrocytes accumulate around the plaques and cause neuroinflammation, where brain chemokines play a major role for the glial accumulation. We have previously shown that transgenic over-expression of chemokine CCL2 in the brain resulted in increased microglial accumulation and diffuse amyloid plaque deposition in a transgenic mouse model of AD expressing Swedish amyloid precursor protein (APP) mutant. Here we report that adeno-associated virus (AAV) serotype 1 and 2 hybrid efficiently deliver 7ND gene, a dominant-negative CCL2 mutant, in dose-response manner and express up to 1000-fold higher amount of recombinant CCL2 over basal level after single administration in brain. AAV1/2 hybrid virus mainly infected neurons without neuroinflammation, and the expression is sustained at least 6-months period. 7ND expressed in APP/presenilin-1 (APP/PS1) bigenic mice reduced astro/microgliosis, beta-amyloidosis, including suppression of both fibrillar and oligomer Aβ accumulation, and improved spatial learning. Our data support that AAV1/2 system is a useful tool for CNS gene delivery, and suppression of CCL2 may be a therapeutic target for the amelioration of AD-related neuroinflammation.
The onset of Alzheimer’s disease (AD) is significantly associated with astro- and microgliosis, which cause brain inflammation 1. Generally, neuroinflammation is beneficial for their neuroprotective role in response to changes in its environment and clearance of damaged tissue or harmful substances. However, under the progressive neurodegeneration, such as AD, Parkinson’s disease and multiple sclerosis, neuroinflammation can also harm neuronal cells through production of cytotoxic molecules (such as pro-inflammatory cytokines, complements, excitotoxins, and reactive oxygen species) (for recent review, see 2). Beta-amyloidosis in AD are modulated by a large number of pro-inflammatory cytokines, and chemokines, such as interferon (IFN)-γ, tumor necrosis factor (TNF)-α, interleukin (IL)-1α/β, IL-6, IL-8, transforming growth factor-β, CD40L and CCL2 (formally monocyte chemotactic protein-1, MCP-1) 2-4. For instance, IFN-γ, in conjunction with TNF-α, has been shown to enhance amyloid precursor protein (APP) mRNA transcription and amyloid-β ( production 5,6 as well as suppression of Aβ degradation 7. APP mice deficient in CD40L or TNFR1 show reduced Aβ deposition, astro/microgliosis 8,9, suggesting that elevated expression of pro-inflammatory cytokines are significantly associated Aβ production in the brain.
CCL2 is a member of the β chemokine subfamily and a signaling ligand as a homodimer for its cognate G-protein coupled receptor CCR2 A/B 10. Activated astrocytes and cells of monocytic origin have been shown to express CCL2 in the brain 11,12. CCL2 has emerged as a mediator of glial activation upon Aβ stimulation and has been detected in senile plaques and reactive microglia 13,14 as well as microvessels 15 in AD brains. To understand the role of CCL2 in AD pathogenesis, we developed a novel animal model (APP/CCL2) 16 by crossing an established Aβ deposition mouse model (Tg2576), which mimics microglial-induced brain inflammation 17,18, with a CCL2 over-expressing mouse (JE-95) 19. We previously showed that CCL2 over-expression within the CNS induces microglial activation and increased diffuse plaque formation. APP/CCL2 mice show early onset of spatial learning impairment, synaptic dysfunction, and enhanced Aβ oligomer formation (manuscript under review). Therefore, the pro-inflammatory cytokines above as well as chemokines such as CCL2 may be potentially therapeutic targets of AD, and our cumulative hypothesis is that increased intrathecal CCL2 is deleterious to APP transgenic mice and that a specific and targeted reduction in intrathecal CNS signaling of CCL2 is beneficial.
To address this hypothesis, we developed a viral gene delivery system to over-express a dominant-negative CCL2 mutant, which is an N-terminal deletion form of CCL2 (7ND) and functions as a CCL2 inhibitor to form a heterodimer with endogenous CCL2 in vitro and in vivo 20,21. We employed an adeno-associated virus (AAV)-mediated gene delivery approach for introducing 7ND. This method has several advantages over other virus-based brain gene delivery systems: (1) minimal immune response compared to adenovirus or herpes simplex virus infections 22; (2) capable of infecting both dividing and non-dividing cells and non-existent spontaneous tumorigenesis as found with retroviral vectors; and (3) persistent transgene expression for more than one year after administration in adult retina 23,24 and brain 25. The AAV1-derived system shows global gene expression in the brain compared to that of AAV2 expressing predominantly in dentate gyrus after intraventricular injection of neonate animal brains 26. We have employed the AAV1/2 hybrid virus system to deliver 7ND to APP/PS1 bigenic mice in this study. Our approach resulted in robust chronic gene expression in CNS, and reduced microglial activation and lower Aβ deposition.
To achieve long-term expression of the dominant negative CCL2 mutant (7ND) in brain, we developed an AAV1/2-mediated somatic gene transfer system. AAV-7ND virus particles that express murine 7ND or AAV-GFP as a control were intracranially injected into the frontal cortex or hippocampus of non-Tg mice, and the total amounts of CCL2 and 7ND were then measured by ELISA.
AAV-7ND injection produces strong 7ND protein expression 2 months after injection in a dose-dependent manner [319-104,551 pg/mg of CCL2/7ND and 3×108-2×1010 viral particles (VP)], whereas injection of control AAV-GFP showed no increase (37-48pg/mg of CCL2 at same doses) (Figure 1a, b), suggesting that most of the 7ND or CCL2 in AAV-7ND injected brain was 7ND recombinant protein. We also examined the time course of 7ND expression from 1 to 24 weeks (Figure 1c). 7ND was expressed at 14,000, 17,300, 70,000, 161,000, and 97,000 pg/mg at 1, 2, 4, 12, and 24 weeks, respectively, after the injection of 2×1010 AAV-7ND VP, whereas control AAV-GFP injected mice expressed 42-394 pg/mg of CCL2 over the same time period, demonstrating long-term expression of recombinant protein. According to the in vitro study, a 10-fold excess amount of 7ND significantly blocks the bioactivity of CCL2 27. Thus, the amount of 7ND recombinant protein is sufficient to inhibit CCL2 protein expressed in the brain of APP/PS1 mice (2,771 +/- 334.3 and 4,026 +/- 781.7 pg/mg brain at 7 and 12 months of age, respectively), a cross breeding of Tg2576 and M146L presenilin-1 familial AD mutant transgenic line 17,28. Interestingly, there was no increase in circulating CCL2 levels in plasma between AAV-7ND and AAV-GFP groups 3 months after intracranial injection (P = 0.7947, Figure 1d). This suggests that the tightly sealed blood brain barrier (BBB) prevents 7ND leakage from the brain parenchyma, and BBB disruption may be necessary for the recruitment of peripheral CCR2+ positive monocytes into brain parenchyma. The AAV-GFP-injected brain was subjected to immunofluorescence to identify AAV-infected cells in brain (Figure 2 a-f). The GFP-positive cells are mainly NeuN-positive, a marker of neurons (a-c), and rarely GFAP-positive, a marker of astrocytes (d-f). The GFP-positive cells are not associated with gliosis (d-f), suggesting the lack of neuroinflammation at 2 months after the gene delivery in this system.
APP/PS1 mice show accelerated Aβ deposition and memory impairment when compared to the original Tg2576 29. Based on our previous results, we hypothesized that treatment of APP/PS1 mice with AAV-7ND can correct Aβ-related pathology in vivo. AAV-7ND and control AAV-GFP viruses were stereotaxically injected into bilateral hippocampal regions of APP/PS1 mice at 3 months of age, with neuropathological analysis at 8 months. Chronic expression of 7ND in APP/PS1 brain reduced astro/microglial accumulation around the β-sheet structure-specific thioflavin-S-positive plaques, which labels β-sheet forming compact plaques, compared to control GFP-expressing APP/PS1 brain (78.1 and 74.2% of AAV-GFP injected APP/PS1 mouse group for astrocytes and microglia, respectively, Figure 3a-j), demonstrating suppression of astro/microgliosis.
Since accumulation of astrocytes and microglia are correlated with amyloid plaque deposition 6,9,30, we examined if 7ND expression suppresses beta-amyloidosis in APP/PS1 mice. First, we evaluated the Aβ42 levels by Aβ42 ELISA as described16. Both SDS-soluble and SDS-insoluble Aβ42 peptides were unchanged by 7ND expression at 8 months of age (Table 1). Next, we examined the area of Aβ loads in the hippocampus of AAV-injected APP/PS1 mice. Beta-amyloidosis is the process of Aβ aggregation from forming oligomers (detected by specific anti-oligomer antibodies) to diffuse plaques (detected by pan anti-Aβ antibody), and eventually compact plaques (thioflavin-S-positive plaques). Although total Aβ load, as determined by anti-Aβ staining, was unchanged by 7ND expression (99.7% of AAV-GFP-injected APP/PS1 mouse group, Figure 4, upper panel, 3.77 +/- 0.86% in GFP and 3.75 +/- 0.27% in 7ND), 7ND significantly reduced thioflavin-S-positive fibrillar Aβ deposition in the hippocampal region (74.4% of AAV-GFP-injected APP/PS1 mouse group, Figure 4, lower panel, 0.40 +/- 0.04% in GFP and 0.30 +/- 0.02% in 7ND). Accumulating evidences suggest that Aβ oligomer is the Aβ aggregate species that mediates memory impairment and neurophysiological dysfunction in transgenic animal models and AD specimens 31,32, and it has been the focus of the characterization of beta-amyloidosis. Quantification of Aβ oligomers by immunohistochemistry using NU-1 and NU-2, anti-oligomeric Aβ antibodies, shows significant reduction of Aβ oligomer load in AAV-7ND injected group as compared to AAV-GFP injected group at 8 months of age (73.8% of AAV-GFP-injected APP/PS1 mouse group, Figure 5a-e, 5.37 +/- 0.26% in GFP and 3.96 +/- 0.11% in 7ND). Dotblot with NU-1 performed for ultracentrifuged brain SDS extracts, which do not contain fibrillar Aβ, also shows significant reduction of Aβ oligomer in the AAV-7ND injected group (Figure 5f). These data suggest that chronic inhibition of CCL2 suppresses astroglial/microglial accumulation and beta-amyloidosis, specifically fibrillar and oligomer species of Aβ aggregates, in APP/PS1 mice.
Finally, we examined if dominant-negative CCL2 improves spatial learning in APP/PS1 mice by 2-day radial arm water maze (RAWM) task (Figure 6) 33. Our initial attempt to determine the effect of pre-symptomatic injection of AAV-7ND on APP/PS1 mice was unsuccessful due to an insignificant difference in spatial learning among non-Tg, APP/PS1 with AAV-7ND, and APP/PS1 with AAV-GFP treatment (data not shown). Thus, we tested the effect of post-symptomatic injection of AAV-7ND in APP/PS1 mice at 10 months of age, with memory function testing at 12 months. The age-matched Non-Tg mouse group modeled the representative learning curve. In contrast, the AAV-GFP-injected APP/PS1 mouse group showed significantly higher errors than that of Non-Tg group throughout the trials, indicating impaired memory acquisition. However, the AAV-7ND-injected APP/PS1 mouse group showed similar error numbers to the Non-Tg group, which is significantly lower than the AAV-GFP group, indicating secured memory acquisition and recall. We also evaluated the Aβ42 levels in the hippocampus of these mouse groups by Aβ42 ELISA (n = 4, Table 1). Although SDS-insoluble Aβ42 peptides were unchanged by 7ND expression, SDS-soluble Aβ42 peptides were significantly reduced by 7ND expression at 12 but not 7 months of age (Table 1). Taken together, these data suggest that chronic inhibition of CCL2 suppresses astrocyte/microglial accumulation, Aβ production and aggregation, and significantly improves memory ability in APP/PS1 mice.
Our results demonstrate that the AAV1/2 hybrid vector is an efficient CNS gene delivery method especially for the expression of a chemokine mutant in the brain due to its high expression level, long-term expression, negligible leak to blood, and minimal inflammatory response after brain parenchymal injection. The AAV1-derived system shows more diffuse gene expression in the brain, whereas the use of the conventional AAV2 vector mainly expresses in the dentate gyrus and with lesser amount of gene expression (data not shown) 26, supporting the use of AAV1/2 hybrid system for more ubiquitous gene delivery in the injected brain region. Using this technology, we have shown that CCL2, which is clinically correlated with AD pathogenesis 16, can be therapeutically targeted. APP/CCL2 mice show an early onset of spatial learning impairment and Aβ oligomer formation, whereas CCL2 mice show normal learning, suggesting that CCL2 is a co-factor in Aβ-induced memory impairment in APP mice but does not induce memory dysfunction by itself (manuscript under review).
The mechanism behind how CCL2 and microgliosis enhance Aβ aggregation is still unknown. Our previous studies ruled out three possibilities: (1) interaction of CCL2 on enhanced Aβ synthesis; (2) reduced clearance of fibrillar Aβ degradation by bone marrow-derived macrophages; and (3) reduced clearance of intracranially injected fibrillar Aβ in vivo 16,34. However, we have recently found that monomeric Aβ undergoes rapid oligomer formation in microglia after uptake, which is enhanced by CCL2 and TNF-α (manuscript under review). This suggests that microglia generate Aβ oligomer intracellularly, which is secreted and serves as an Aβ oligomer seed for diffuse plaque formation. Thus, blockage of CCL2 by the expression of 7ND may suppress beta-amyloidosis via inhibition of microgliosis and direct suppression of intracellular Aβ oligomer formation in microglia.
Disruption of CCR2, a specific CCL2 receptor, inhibited microglial accumulation in APP mice, which is consistent with our findings 35. However, APP/CCR2-/- mice show early onset of amyloid deposition 35. This could be due to the lack of infiltration by peripheral bone marrow-derived monocytes into the brain 36. The lack of peripheral monocytes, which play a key role in amyloid clearance, may lead to enhanced Aβ deposition in the brain. Since most AD patients have intact CCL2-CCR2 signaling, suppression of chemokine-mediated mononuclear phagocyte flux via chemokine inhibitors may be of therapeutic interest in the future. Other chemokines, such as IL-8, upregulated in MCI and AD patients, also may have similar functions in microgliosis and cognitive dysfunction in AD mouse models. The functions of these other chemokines could be investigated in the future.
Adenoviral gene delivery of 7ND has been used for the treatment of multiple inflammatory disease animal models in vivo, while there is only one report of the adenoviral 7ND gene transfer into animal brain 37 and there were no studies of the AAV-mediated gene transfer of 7ND. We found no difference in the longevity of AAV-7ND injected APP/PS1 mice as compared to AAV-GFP injected group, socio-behavioral activities (diet, sleep, spontaneous activities, motor coordination, aggressiveness, irritation, cleanness, fatigue, etc), or body weight. AAV-mediated gene delivery has been successful in reducing beta-amyloidosis in mouse brains by the expression of anti-Aβ single chain variant fragments 38,39, Aβ degrading enzyme (neprilysin and endothelin converting enzyme) 40,41, recombinant Aβ fusion protein for vaccination 42-44, and integral membrane protein 2B familial British dementia mutant protein (BRI2) 45. AAV-mediated Aβ vaccination of APP mice also improved working memory 42,44. To the best of our knowledge, this is the first time that AAV-mediated gene delivery of a chemokine antagonist into transgenic AD animal models has been used to suppress beta-amyloidosis and improve working memory in APP/PS1 mice.
Although non-steroidal anti-inflammatory drugs (NSAIDs) have been proposed as prophylactic drugs 46 or drugs capable of ameliorating beta-amyloidosis in transgenic amyloid mouse models, none of them were successful in clinical trials except R-flurbiprofen 47. In addition, NSAIDs have been the subjects of chemical engineering for their gamma-secretase modulating activity instead of their original anti-inflammatory activity 48. Thus, both AAV-mediated anti-inflammatory therapy and gamma-secretase modulating drugs may be mutually synergistic for the treatment of beta-amyloidosis.
In summary, our study demonstrates the potential of AAV1/2 hybrid-mediated CNS gene delivery of chemokine mutants, which may be beneficial to suppress the neuroinflammation and beta-amyloidosis, which are detrimental to neuronal function. Further study will determine how a blockade of brain CCL2 could affect clinical AD progression.
Tg2576 mice expressing the Swedish mutation of human APP695 were obtained from Drs. G. Carlson and K. Hsiao-Ashe through Mayo Medical Venture 17. Male Tg2576 were backcrossed at least seven times to the B6/129 strain. PS1 mutant mice (M146L line 6.1) were provided by Dr. K. Duff through University of South Florida, maintained as PS1 transgene homozygote28, which have been back-crossed to the B6/129 strain at least five times prior to the crossing. To generate APP/PS1 bigenic mice, Tg2576 mice were crossed with PS1 mice. Age-matched non-tg mice in B6/129 F1 strain (Jackson laboratory, Bar Harbor, ME) were maintained by intercrossing in the same facility. All animal use procedures were strictly reviewed by Institutional Animal Care and Use Committee (IACUC) of University of Nebraska Medical Center.
We have constructed an AAV2 vector (pAAV2-MCS-WPRE) which contains cytomegalovirus (CMV) immediate early enhancer, chicken β-actin promoter with first exon and intron sequences, multiple cloning site (MCS), Woodchuck hepatitis post-transcriptional regulatory element (WPRE), and the bovine growth hormone polyadenylation site, all of which are flanked by AAV2 inverted terminal repeat based on pGFP vector (provided by R. Klein) 49. To construct 7ND, mouse CCL2 was generated by proof-reading PCR amplification of wild type CCL2 (1516 bp) from mouse genomic DNA and subcloned into the MCS of pUC18 at EcoRI-BamHI sites. After sequencing, the first 7 amino acids after the signal peptide sequence (25PDAVNAP31) were deleted by site-directed mutagenesis. Next, the entire 7ND genomic DNA was subcloned into MCS of pAAV-MCS-WPRE at EcoRI-NheI sites, developing pAAV-MCS-WPRE-7ND. For details of primer sets used for construction of AAV-7ND and generation and purification of AAV1/2 virus, see supplementary material.
Mice at 3 months of age received i.p. injection of ketamine/xylazine anesthesia (100mg/kg ketamine and 20 mg/kg xylazine). After mice were immobilized in a stereotaxic apparatus (Stoelting, Wood Dale, IL), a linear skin incision was made over the bregma, and a 1-mm burr hole was drilled in the skull 2.1 mm posterior and 1.8 mm lateral to the bregma on both sides using a hand-held driller. 2 μl of saline containing AAV-GFP or AAV-7ND (2 × 1010 VP) was injected into hippocampus 1.8 mm below the surface of the skull using 10-μl Hamilton syringe.
Immunohistochemistry was performed as described previously 6,16. Briefly, mice were euthanized with isoflurane and perfused transcardially with 25 ml of ice cold PBS. The brains were rapidly removed, immersed in freshly depolymerized 4% paraformaldehyde for 48 hours, and cryoprotected by successive 24-hour immersions in 15% and 30% sucrose in 1× PBS immediately before sectioning. Fixed, cryoprotected brains were frozen and sectioned in the horizontal plane using a Cryostat (Leica, Bannockburn, IL), with sections collected serially. Immunohistochemistry was performed using specific antibodies to identify astrocyte (GFAP, rabbit polyclonal, 1:2000)(DAKO, Denmark), microglia (IBA1, rabbit polyclonal, 1:1000)(Wako, Richmond, VA), microtubule-associated protein 2 (MAP2) (Chemicon, Temecula, CA), pan-Aβ (1:100)(Zymed, San Francisco, CA) and Aβ oligomer (NU-1 or NU-2 monoclonal, 1μg/ml)(Northwestern University). Immunodetection was visualized using Envision Plus (DAKO, Carpinteria, CA) with 3,3′-diaminobenzidine (Vector Laboratories, Burlingame, CA) for Aβ oligomer staining. For immunofluorescence, Alexa Fluor®568-conjugated anti-rabbit IgG (H+L) was used as a secondary antibody. 0.1% Thioflavin S in 50% EtOH was used for counterstaining of compact plaque (Sigma, St. Louis, MO). For quantification analysis, the numbers of GFAP-positive astrocytes and IBA1-positive microglia around Aβ plaques in the hippocampus were counted at 100-μm intervals in ten 10-μm coronal sections from each mouse. Three TS-positive plaques were randomly chosen per section, and five brains of mice per group were analyzed (30 plaques per animal). The areas of Aβ loads and TS-positive plaques were analyzed using an image analysis software (ImageJ, NIH) at 100-μm intervals in ten 10-μm coronal sections from each mouse. Five brains of mice per group were analyzed.
Aβ and CCL2 ELISA were performed as previously described 6,16,50. Dot-blot analysis for Aβ oligomer was performed as described previously 50 with some modifications. Briefly, mouse brains were homogenized in 2% SDS containing PMSF (0.5mM), Leupeptin (3μg/ml) and Aprotinin (5μg/ml), and centrifuged at 4 °C for 1 h at 100,000 × g. The supernatant was used for dot blot. Protein concentration was determined using BCA Protein Assay Kit (Pierce, Rockford, IL). Samples were adjusted with 1× PBS to the same concentration, and applied to a nitrocellulose membrane using Bio-Dot® SF Microfiltration Apparatus according to manufacturer’s instruction (Bio-Rad, Richmond, CA). After blotting, membrane was incubated in a 5% solution of nonfat dry milk for 1 h at room temperature. After overnight incubation at 4 °C with primary antibody (clone NU-1; 1μg/ml), membrane was washed in Tween 20-TBS (TTBS) (0.05% Tween 20, 100 mM Tris, pH 7.5, 150 nM NaCl) for 3× 10 min and incubated at room temperature with the HRP-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) for 1 h. Membrane was washed in TTBS for 3×10 min and incubated for 5 min with chemiluminescent substrate (ECL Plus, GE Healthcare Bio-Sciences, Italy). Images were taken using an image analyzer (Typhoon, Amersham Biosciences, Piscataway, NJ). Band density was measured using ImageQuant software.
The radial arm water maze task was run as described previously with minor modification 33. Animals were introduced into the perimeter of a circular water-filled tank 110 cm in diameter and 91 cm in height (San Diego Instruments, San Diego, CA) with triangular inserts placed in the tank to produce six swim paths radiating out from a central area. Spatial cues for mice orientation were present on the walls of the tank. At the end of one arm, a 10 cm circular plexiglass platform was submerged 1 cm deep—hidden from the mice. On day one, 15 trials (12 trials with visible platform followed by 3 trials with hidden platform) were run in five blocks of 3. A cohort of 4 mice was run sequentially for each block. After each 3-trial block, a second cohort of mice was run permitting an extended rest period before mice were exposed to the second block. The target arm location remained constant for a given mouse throughout the test. Each trial lasts 1 min and an error is scored each time the body of the mouse, excluding tail, enters the wrong arm, enters the arm with the platform but does not climb on it, or does not make a choice for 20 s. Each trial ends when the mouse climbs onto and remains on the hidden platform for 10 seconds. The mouse is given 20 s to rest on the platform between each trial. On day two, the mice were run in exactly the same manner as day one except that the platform was hidden for all trials. Between blocks 4 and 5 on day 1, and blocks 9 and 10 on day two, additional retention for 30 minutes was given to each mouse to test short-term memory recall. The errors on each block were averaged and used for statistical analysis.
All data were normally distributed and presented as mean values ± standard errors of mean (SEM). In case of single mean comparison, data were analyzed by Student’s t-test. In case of multiple mean comparisons, the data were analyzed by two-way repeated measures analysis of variance (ANOVA), followed by Bonferroni multiple comparison test using statistics software (Prism 4.0, Graphpad Software, Inc., San Diego, CA). p values less than 0.05 were regarded as significant difference.
We thank Drs. K. Hsiao-Ashe for providing Tg2576 mice, K. Duff for providing M146L PS1 mice, D. Morgan for RAWM test training and consultation, R. Klein for pGFP plasmid, and University of Pennsylvania Gene Therapy Program for recombinant AAV1 vectors, and Meg Marquardt for editing of the manuscript. This work is supported by Vada Oldfield Alzheimer Research Foundation (TI, TK), NIH P01 NS043985 (TI, HEG), and NIH R01 NS32151 (RMR).